manual: add extracted figures to qmc statistics lab
git-svn-id: https://subversion.assembla.com/svn/qmcdev/trunk@6881 e5b18d87-469d-4833-9cc0-8cdfa06e9491
After Width: | Height: | Size: 4.2 KiB |
After Width: | Height: | Size: 4.4 KiB |
After Width: | Height: | Size: 3.7 KiB |
After Width: | Height: | Size: 3.1 KiB |
After Width: | Height: | Size: 3.5 KiB |
After Width: | Height: | Size: 44 KiB |
After Width: | Height: | Size: 46 KiB |
|
@ -429,6 +429,14 @@ behavior. qmca plots traces with the -t flag.
|
|||
Type \textbf{qmca -q e -t H.s000.scalar.dat}, which produces a graph of the
|
||||
trace of the local energy:
|
||||
|
||||
\FloatBarrier
|
||||
\begin{figure}[ht!]
|
||||
\begin{center}
|
||||
\includegraphics[trim = 0mm 0mm 0mm 0mm, clip,width=0.75\columnwidth]{./figures/lab_qmc_statistics_tracing1}
|
||||
\end{center}
|
||||
\end{figure}
|
||||
\FloatBarrier
|
||||
|
||||
%\includegraphics[scale=0.5]{E_L_H_STO-2G.png}
|
||||
|
||||
The solid black line connects the values of the local energy at each MC block
|
||||
|
@ -445,6 +453,14 @@ comparable to the total local energy.
|
|||
Change to directory \texttt{problematic} and type \textbf{qmca -q e -t
|
||||
H.s000.scalar.dat} to produce this graph:
|
||||
|
||||
\FloatBarrier
|
||||
\begin{figure}[ht!]
|
||||
\begin{center}
|
||||
\includegraphics[trim = 0mm 0mm 0mm 0mm, clip,width=0.75\columnwidth]{./figures/lab_qmc_statistics_tracing2}
|
||||
\end{center}
|
||||
\end{figure}
|
||||
\FloatBarrier
|
||||
|
||||
%\includegraphics[scale=0.5]{E_L_H_B-splines.png}
|
||||
|
||||
Here, the local energy samples cluster around the expected -0.5 hartrees for the
|
||||
|
@ -505,6 +521,14 @@ see this, type \textbf{qmca -q e {-}{-}sac *.scalar.dat} to see the energies
|
|||
and autocorrelation times, then plot with gnuplot by inputting \textbf{gnuplot
|
||||
H.plt}:
|
||||
|
||||
\FloatBarrier
|
||||
\begin{figure}[ht!]
|
||||
\begin{center}
|
||||
\includegraphics[trim = 0mm 0mm 0mm 0mm, clip,width=0.75\columnwidth]{./figures/lab_qmc_statistics_blocking1}
|
||||
\end{center}
|
||||
\end{figure}
|
||||
\FloatBarrier
|
||||
|
||||
%\includegraphics[scale=1.0]{timestep_vs_autocorrelation_energy_H_STO-2G.png}
|
||||
|
||||
The error bar also increases with the autocorrelation.
|
||||
|
@ -550,6 +574,15 @@ visualize the data:
|
|||
%H series 15 LocalEnergy = -0.465723 +/- 0.004425 2.6
|
||||
%\end{verbatim}
|
||||
%\end{shaded}
|
||||
|
||||
\FloatBarrier
|
||||
\begin{figure}[ht!]
|
||||
\begin{center}
|
||||
\includegraphics[trim = 0mm 0mm 0mm 0mm, clip,width=0.75\columnwidth]{./figures/lab_qmc_statistics_blocking2}
|
||||
\end{center}
|
||||
\end{figure}
|
||||
\FloatBarrier
|
||||
|
||||
%\includegraphics[scale=1.0]{steps_per_block_vs_autocorrelation_energy_H_STO-2G.png}
|
||||
|
||||
The greatest number of steps per block produces the smallest autocorrelation
|
||||
|
@ -651,6 +684,14 @@ factors of four from 32 to 128 to 512. Type \textbf{qmca -q ev *scalar.dat}
|
|||
and note the change in the error bar on the local energy as the number of
|
||||
nodes. Visualize this with \textbf{gnuplot H.plt}:
|
||||
|
||||
\FloatBarrier
|
||||
\begin{figure}[ht!]
|
||||
\begin{center}
|
||||
\includegraphics[trim = 0mm 0mm 0mm 0mm, clip,width=0.75\columnwidth]{./figures/lab_qmc_statistics_nodes}
|
||||
\end{center}
|
||||
\end{figure}
|
||||
\FloatBarrier
|
||||
|
||||
%\includegraphics[scale=1.0]{nnode_vs_energy_H_STO-2G.png}
|
||||
|
||||
Increasing the number of blocks, unlike running in parallel, increases the
|
||||
|
@ -664,6 +705,14 @@ To see the effect of increasing the block number, change to the directory
|
|||
increasing the total number of samples. Visualize the tradeoff by inputting
|
||||
\textbf{gnuplot H.plt}:
|
||||
|
||||
\FloatBarrier
|
||||
\begin{figure}[ht!]
|
||||
\begin{center}
|
||||
\includegraphics[trim = 0mm 0mm 0mm 0mm, clip,width=0.75\columnwidth]{./figures/lab_qmc_statistics_blocks}
|
||||
\end{center}
|
||||
\end{figure}
|
||||
\FloatBarrier
|
||||
|
||||
%\includegraphics[scale=1.0]{nblock_vs_tcpu_energy_H_STO-2G.png}
|
||||
|
||||
Press \textbf{q [Enter]} to quit gnuplot.
|
||||
|
@ -760,6 +809,14 @@ Use \textbf{qmca -q bc *scalar.dat} to see that the CPU time per block
|
|||
increases with number of electrons in the simulation, then plot the total CPU
|
||||
time of the simulation by \textbf{gnuplot Nelectron\_tCPU.plt}:
|
||||
|
||||
\FloatBarrier
|
||||
\begin{figure}[ht!]
|
||||
\begin{center}
|
||||
\includegraphics[trim = 0mm 0mm 0mm 0mm, clip,width=0.75\columnwidth]{./figures/lab_qmc_statistics_scaling}
|
||||
\end{center}
|
||||
\end{figure}
|
||||
\FloatBarrier
|
||||
|
||||
%\includegraphics[scale=1.0]{nelectron_vs_tcpu_H_C_CH_STO-6G.png}
|
||||
|
||||
The green pluses represent the CPU time per block at each electron number.
|
||||
|
|